def load_weights():
pretrained_path = 'BioBertFolder/biobert_v1.0_pubmed_pmc/'
# ffn_weight_file = None
bert_ffn_weight_file = 'newFolder/models/bertffn_crossentropy/bertffn'
embedding_file = 'Float16EmbeddingsExpanded5-27-19.pkl'
doc = RetreiveQADoc(pretrained_path=pretrained_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_ffn_weight_file,
embedding_file=embedding_file)
var = 'yes'
while (var == 'yes'):
inp = input("How are you feeling today?: ")
print(doc.predict(inp, search_by='answer', topk=1, answer_only=True))
var = input("Do you want more assistance?")
def get_prediction(query):
'''
Will return the analysed things after predicting from the query
'''
PRETRAINED_PATH = os.path.join(os.path.expanduser("~"), "data", "BioBert", "biobert_v1.0_pubmed_pmc")
BERT_FFN_WEIGHT_FILE = os.path.join(os.path.expanduser("~"), "data", "newFolder", "models", "bertffn_crossentropy", "bertffn")
EMBEDDING_FILE = os.path.join(os.path.expanduser("~"), "data", "Float16EmbeddingsExpanded5-27-19.pkl")
DOCTOR = RetreiveQADoc(
pretrained_path=PRETRAINED_PATH,
ffn_weight_file=None,
bert_ffn_weight_file=BERT_FFN_WEIGHT_FILE,
embedding_file=EMBEDDING_FILE
)
return DOCTOR.predict(
query,
search_by="answer",
topk=1,
answer_only=True)
class QAModel:
def __init__(self, biobert_path: str, bert_fnn_weights: str,
embedding_file: str):
self._qa_model = RetreiveQADoc(pretrained_path=biobert_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_fnn_weights,
embedding_file=embedding_file)
def predict(self,
question_text,
search_by: str = 'answer',
results_num: int = 1,
answer_only: bool = True):
results = self._qa_model.predict(question_text,
search_by=search_by,
topk=results_num,
answer_only=answer_only)
return results
import zipfile
zip_ref = zipfile.ZipFile('DataAndCheckpoint.zip', 'r')
zip_ref.extractall('newFolder')
zip_ref.close()
#@title Load model weights and Q&A data. Double click to see code
from docproduct.predictor import RetreiveQADoc
pretrained_path = 'BioBertFolder/biobert_v1.0_pubmed_pmc/'
# ffn_weight_file = None
bert_ffn_weight_file = 'newFolder/models/bertffn_crossentropy/bertffn'
embedding_file = 'Float16EmbeddingsExpanded5-27-19.pkl'
doc = RetreiveQADoc(pretrained_path=pretrained_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_ffn_weight_file,
embedding_file=embedding_file)
''' predict on a single sentence just for the testing purpose. '''
test_sen1 = [0,0]
test_sen1[0] = "This is one of the best creation of Nolan. I can say, it's his magnum opus. Loved the soundtrack and especially those creative dialogues."
test_sen1[1] = "I have fever for past few days."
for t in test_sen1:
test_sen1 = TEXT.preprocess(t)
test_sen1 = [[TEXT.vocab.stoi[x] for x in test_sen1]]
test_sen = np.asarray(test_sen1)
test_sen = torch.LongTensor(test_sen)
test_tensor = Variable(test_sen, volatile=True)
def __init__(self, biobert_path: str, bert_fnn_weights: str,
embedding_file: str):
self._qa_model = RetreiveQADoc(pretrained_path=biobert_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_fnn_weights,
embedding_file=embedding_file)
from docproduct.predictor import RetreiveQADoc
pretrained_path = 'BioBertFolder/biobert_v1.0_pubmed_pmc/'
# ffn_weight_file = None
bert_ffn_weight_file = 'newFolder/models/bertffn_crossentropy/bertffn'
embedding_file = 'Float16EmbeddingsExpanded5-27-19.pkl'
doc = RetreiveQADoc(pretrained_path=pretrained_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_ffn_weight_file,
embedding_file=embedding_file)
question_text = "i am 18 and having frequent and painful urination, with a mild fever since 2-3 days"
search_similarity_by = 'answer'
number_results_toReturn = 10
answer_only = True
returned_results = doc.predict(question_text,
search_by=search_similarity_by, topk=number_results_toReturn, answer_only=answer_only)
print('')
for jk in range(len(returned_results)):
print("Result ", jk+1)
print(returned_results[jk])
print('')
from docproduct.predictor import RetreiveQADoc
pretrained_path = 'pubmed_pmc_470k/'
ffn_weight_file = None
bert_ffn_weight_file = 'models/bertffn_crossentropy/bertffn'
embedding_file = 'qa_embeddings/bertffn_crossentropy.pkl'
doc = RetreiveQADoc(pretrained_path=pretrained_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_ffn_weight_file,
embedding_file=embedding_file)
print(
doc.predict('my eyes hurts and i have a headache.',
search_by='answer',
topk=5,
answer_only=True))
print(
doc.predict('my eyes hurts and i have a headache.',
search_by='question',
topk=5,
answer_only=True))
from flask import Flask, jsonify, make_response, request
app = Flask(__name__)
MODEL_LOADED = False
REPO_LOCATION = os.path.join("code", "Hackshetra", "py_backend")
PRETRAINED_PATH = os.path.join(os.path.expanduser("~"), REPO_LOCATION, "data",
"BioBert", "biobert_v1.0_pubmed_pmc")
BERT_FFN_WEIGHT_FILE = os.path.join(os.path.expanduser("~"), REPO_LOCATION,
"data", "newFolder", "models",
"bertffn_crossentropy", "bertffn")
EMBEDDING_FILE = os.path.join(os.path.expanduser("~"), REPO_LOCATION, "data",
"Float16EmbeddingsExpanded5-27-19.pkl")
DOCTOR = RetreiveQADoc(pretrained_path=PRETRAINED_PATH,
ffn_weight_file=None,
bert_ffn_weight_file=BERT_FFN_WEIGHT_FILE,
embedding_file=EMBEDDING_FILE)
MODEL_LOADED = True
print("\n\n MODEL LOADED YAY 😀")
@app.route('/wiki', methods=['POST'])
def table():
'''
Will return the data from side table from WikiPedia
'''
try:
query = request.json['query']
return jsonify(wiki.wiki_get_table(query))
tar.extractall(path='BioBertFolder/')
tar.close()
file_id = '1uCXv6mQkFfpw5txGnVCsl93Db7t5Z2mp'
download_file_from_google_drive(file_id,
'Float16EmbeddingsExpanded5-27-19.pkl')
file_id = 'https://onedrive.live.com/download?cid=9DEDF3C1E2D7E77F&resid=9DEDF3C1E2D7E77F%2132792&authkey=AEQ8GtkcDbe3K98'
urllib.request.urlretrieve(file_id, 'DataAndCheckpoint.zip')
if not os.path.exists('newFolder'):
os.makedirs('newFolder')
zip_ref = zipfile.ZipFile('DataAndCheckpoint.zip', 'r')
zip_ref.extractall('newFolder')
zip_ref.close()
from docproduct.predictor import RetreiveQADoc
pretrained_path = 'BioBertFolder/biobert_v1.0_pubmed_pmc/'
# ffn_weight_file = None
bert_ffn_weight_file = 'newFolder/models/bertffn_crossentropy/bertffn'
embedding_file = 'Float16EmbeddingsExpanded5-27-19.pkl'
doc = RetreiveQADoc(pretrained_path=pretrained_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_ffn_weight_file,
embedding_file=embedding_file)
def load_model():
global doc
doc = RetreiveQADoc(pretrained_path=pretrained_path,
ffn_weight_file=None,
bert_ffn_weight_file=bert_ffn_weight_file,
embedding_file=embedding_file)